Nearest Neighbor Interaction PSO Based on Small-World Model

Abstract

Particle swarm optimization with passive congregation (PSOPC) is a novel variant of particle swarm optimization (PSO) by simulating the animal congregation phenomenon. Although it is superior to the standard version in some cases, however, due to the randomly selected neighbor particle, the performance of PSOPC is not always stable. Therefore, in this paper, a new variant – nearest neighbor interaction particle swarm optimization based on small world model (NNISW) is designed to solve this problem. In NNISW, the additional congregation item is associated with the best particle, nor the random ones, and the small world topology structure is introduced also to simulate the true swarm behavior. After compared with other seven famous benchmarks in high-dimensional cases, the performance of this new variant is superior to other three variants of PSO.Keywordsparticle swarm optimizationsmall-worldcongregation